Time  Nick       Message
00:01 paramat    will merge game#2282 after i test it doesn't break anything
00:02 ShadowBot  https://github.com/minetest/minetest_game/issues/2282 -- Remove uses of default.gui_bg/bg_img/slots by paramat
01:55 paramat    tested, merging game 2282
01:57 paramat    done
08:09 oiaohm     src/noise.cpp:432:29: warning: ‘void* memcpy(void*, const void*, size_t)’ writing to an object of type ‘struct NoiseParams’ with no trivial copy-assignment; use copy-assignment or copy-initialization instead [-Wclass-memaccess] << has this fault been fixed elsewhere in the code before.
11:10 nerzhul    merged #8000
11:10 ShadowBot  https://github.com/minetest/minetest/issues/8000 -- CSM restrictions: Make 'LOAD_CLIENT_MODS' disable loading of 'builtin' by paramat
11:11 nerzhul    #7577 has been fixed by reverting the bad commit
11:11 ShadowBot  https://github.com/minetest/minetest/issues/7577 -- [0.4.17] 'upright_sprite' players are positioned 1 node above ground
11:39 Krock      nerzhul: Fix what? https://github.com/minetest/minetest/pull/8030#issuecomment-451113975
11:40 nerzhul    https://github.com/minetest/minetest/pull/8030#discussion_r244504955 :)
11:40 nerzhul    I'm just generating a minetest 0.4.17.3 release for Android i hope some can test it
11:47 Krock      nerzhul: done
11:48 Krock      updated #8015 with the correct offset calculation
11:48 ShadowBot  https://github.com/minetest/minetest/issues/8015 -- Camera: Code improvements, use Raycast by SmallJoker
11:49 Krock      ^ visual improvement but can wait until 5.1
11:50 Krock      also relevant: anticheat and object properties getter fixes
11:50 Krock      #8013
11:50 ShadowBot  https://github.com/minetest/minetest/issues/8013 -- Fix various bugs (Anticheat, Lua helpers) by SmallJoker
11:51 nerzhul    https://github.com/minetest/minetest/releases/download/0.4.17.1/Minetest-0.4.17.3.apk
11:51 nerzhul    new apk to test to puiblish on the store
11:51 Krock      thanks for the build :)
11:51 nerzhul    gui scaling not backported yet but i can try to backport it after
11:51 nerzhul    it works as intended on my phone
11:51 Krock      5.0 will solve the gui scaling too
11:51 nerzhul    (huawei honor 10, android 6)
11:51 nerzhul    yes if we release 5.0 in 1 month can be nice
11:52 nerzhul    i don't understand why we don't just triggered the freeze
11:52 nerzhul    remaining PR are polishing PR
11:52 nerzhul    it can be included
11:52 nerzhul    i will publish this APK in the beta on the play store
11:52 Krock      well, somebody needs to update the topic :)
11:52 Krock      (IRC topic)
11:53 nerzhul    we should release at the end of the month
11:53 nerzhul    we need more return
11:53 nerzhul    (and if too many bugs we can delay to polish the release) but we should have a defined target
11:53 nerzhul    it seems we are three active coredev , you, paramat and me
11:53 nerzhul    oh and rubenwardy :)
11:53 nerzhul    4
11:56 nerzhul    interesting 300 installations are in the beta program on the play store (over 7k5 installations)
11:56 p_gimeno   and Shara, but mostly on MTG
12:00 Shara      I only really do MTG, and documentation
12:08 Krock      well, ruben still has a few issues open
12:09 Krock      but he doesn't have time recently
12:18 rubenwardy I'm waiting for pr reviews on one
12:18 rubenwardy And then I have another 2 prs to make
12:18 rubenwardy Then testing
12:20 Krock      which one needs reviewing?
12:20 rubenwardy #7892
12:20 ShadowBot  https://github.com/minetest/minetest/issues/7892 -- Deprecate modpack.txt and use modpack.conf instead by rubenwardy
12:21 rubenwardy I'm starting to wish we had a more generic content.conf or similar
12:21 rubenwardy It would make testing for type more complex than a file check though
12:21 rubenwardy Also
12:21 rubenwardy Texture_pack.conf and modpack.conf
12:22 rubenwardy Argh
12:24 rubenwardy One of those PRs I could make now, probably
12:24 Krock      testing...
12:25 Krock      rubenwardy: MPs are no longer grouped
12:25 Krock      only the modpack name is shown in the config dialogue
12:28 rubenwardy I swear, I fix one thing and something else breaks
12:30 Krock      generic code is somewhat non-existent in th emainmenu
12:32 rubenwardy I'd like to move all of the package and content store code into the core at some point to reduce all this duplication
12:33 rubenwardy Well obviously not the ui
12:45 nerzhul    rubewardy is that breaking changes ?
12:46 nerzhul    or changes with compat ?
12:46 Krock      compat changes
12:47 nerzhul    then we can delay it if needed. I hope we can do a three month or four month dev cycle with less changes in next cycles as now we have fixed most old code parts which needs breakages
12:47 nerzhul    no ?
13:04 Krock      well, some of the PRs are required so that the content store works correctly
13:08 nerzhul    Krock: just added my review on the float pr
13:08 nerzhul    Okay noted, and nobody can help to make them being mergeable ?
13:09 Krock      well, I don't understand the mainmenu code
13:12 nerzhul    it's very old and some parts are not cool yeah
13:12 nerzhul    i just reviewed #7892
13:12 ShadowBot  https://github.com/minetest/minetest/issues/7892 -- Deprecate modpack.txt and use modpack.conf instead by rubenwardy
13:12 nerzhul    i think lua for a such menu is a little bit complicated
13:13 rubenwardy definitely
13:13 Krock      nerzhul: comments addressed.
13:13 nerzhul    Krock ty :)
13:15 nerzhul    then we should have same problem of precision in map format, at least it's less visible :p
13:15 nerzhul    every F1000 is used ? none can be dropped ?
13:16 Krock      yes
13:16 Krock      well readV2F1000 is only used in the unittests but I'm not keen changing these in this PR
13:17 nerzhul    why ? do you want to use it later ?
13:17 nerzhul    if it's dead code drop, we have git commits to retrieve it
13:17 nerzhul    one less historical problem in the code :)
13:33 nerzhul    i just updated changelog from 5aa865e56b18c3f0cf1d43da2298ccc05236e15c to 6b102ce51f01a381ab6356931689d85528455a50
13:34 nerzhul    oh we should add the player backend depreciation too
13:34 nerzhul    if i remember files is always the default and we said in this release we must use sqlite3
13:35 nerzhul    and we should add auth backend depreciation too
13:35 nerzhul    and i forget to add postgresql support on auth...
13:35 nerzhul    it's the most interesting auth part :p
13:54 Krock      but those features don't really have priority for 5.0.0
14:00 Krock      nerzhul: removed V2F1000 methods, updated & tested #8030
14:00 ShadowBot  https://github.com/minetest/minetest/issues/8030 -- Proselytize the network. Use IEEE F32 by SmallJoker
14:01 nerzhul    src/unittest/test_world/world.mt
14:01 nerzhul    you pushed test_world :p
14:01 nerzhul    i just updated https://wiki.minetest.net/index.php?title=Database_backends
14:01 nerzhul    to add auth backend
14:03 nerzhul    i just pushed #8043 for 5.0.0 too, we forgot it
14:03 Krock      oof
14:03 ShadowBot  https://github.com/minetest/minetest/issues/8043 -- Make sqlite3 default auth & player backends for new worlds by nerzhul
14:03 nerzhul    make sqlite3 the default for player & auth backends instead of files, keeping compact with old non migrated worlds
14:03 nerzhul    compat*
14:05 Krock      gitignore: src/unittest/test_world/world.mt
14:05 Krock      but yet it is not ignored
14:06 nerzhul    yeah because it has some unittest data
14:06 nerzhul    it's not a totally ignored file :p
14:06 nerzhul    8043 is not complete
14:06 nerzhul    i need to fix the succeded algorithm because file was saved and then re-read :p
14:11 nerzhul    okay #8043 is ready
14:11 ShadowBot  https://github.com/minetest/minetest/issues/8043 -- Make sqlite3 default auth & player backends for new worlds by nerzhul
14:11 nerzhul    tested it in my old worlds )
14:17 Krock      LGTM, didn't test
14:21 nerzhul    better to use split than unified to review
14:22 nerzhul    if you want to test compile start with your world, ou should have a warning for auth :p
14:22 nerzhul    new world is sqlite
14:31 nerzhul    Krock waiting for CI but you change seems good to me
14:52 p_gimeno   [0103 15:01:09] <nerzhul> you pushed test_world :p
14:53 p_gimeno   I've had problems with that too, could the test be made so that test_world is reverted or automatically generated?
14:54 Krock      automatically *deleted
14:55 p_gimeno   automatically generated by the test, so that it doesn't need to be in the repo in the first place
14:56 p_gimeno   possibly automatically deleted, but that's not really important since it's already .gitignore'd
14:56 nerzhul    you can add it to git ignore
14:57 nerzhul    but if it's modified remember you should do git add -f on the file
14:57 nerzhul    there are default values if i remember
14:57 p_gimeno   the tests modify it automatically, so every time you run a test you have to remember to do that
14:58 p_gimeno   it's a bit annoying, the test could just generate it at run time instead
14:59 p_gimeno   or even have a template e.g. world.mt.template, and have the test copy it to world.mt
14:59 nerzhul    yep maybe it can be a nice idea
15:00 p_gimeno   that's not a blocker though, just a minor annoyance
15:00 p_gimeno   it can wait to after 5.0.0
15:44 nerzhul    and it's tests, yes if we got time before we can fix, but it doesn't affect the code, just dev life :)
15:53 nerzhul    Krock: when you get time, if you can test/approve #8043
15:53 ShadowBot  https://github.com/minetest/minetest/issues/8043 -- Make sqlite3 default auth & player backends for new worlds by nerzhul
15:54 rubenwardy sqlite player backends result in massive performance issues
15:55 nerzhul    in which direction ?
15:56 nerzhul    it should be far more better than regular files on large servers
15:57 nerzhul    i used postgresql instead of sqlite and the performance is very good, the penalty on this backend is not visible
15:57 nerzhul    (15k accounts)
16:04 nerzhul    tested #8030 , merging
16:04 ShadowBot  https://github.com/minetest/minetest/issues/8030 -- Proselytize the network. Use IEEE F32 by SmallJoker
16:04 nerzhul    (and approved)
16:04 nerzhul    ty Krock
16:06 nerzhul    rubewardy: as we don't sync and flush the sqlite3 on each tick for player backends i don't see the point there
16:06 nerzhul    and for huge servers i suggest to use postgresql instead of sqlite3, but the sqlite3 contension on player should be ridiculous compared to sqlite3 map
16:15 rubenwardy nerzhul: I don't see why that would be true, given that maps can store thousands of blocks
16:15 rubenwardy and sauth too
16:15 rubenwardy without locking up the server for seconds
17:29 Krock      nerzhul: thanks for the approval & merge. Will test your sqlite3 change probably tomorrow
18:53 p_gimeno   nerzhul: unit tests added to #8019
18:53 ShadowBot  https://github.com/minetest/minetest/issues/8019 -- Use true pitch/yaw/roll rotations without loss of precision by pgimeno by pauloue
20:43 sofar      #8044 for 5.0, remove essentially obsolete code
20:43 ShadowBot  https://github.com/minetest/minetest/issues/8044 -- Remove remote media compatibility mode. by sofar
20:43 sofar      or 5.1, idc really
22:12 nerzhul    rubenwardy, then it seems we agree that sqlite3 has no problem :p
22:14 nerzhul    sofar interesting change, i never saw that code part, how did your found it ?
22:16 nerzhul    p_gimeno: nice
22:18 sofar      nerzhul: well I run the PRMS, so, that sort of behavior was easy to spot
22:18 nerzhul    PRMS ?
22:18 sofar      Public Remote Media Server
22:19 nerzhul    oh it's used ?
22:19 sofar      yes, quite a bit
22:19 nerzhul    how do you make it work ?
22:19 nerzhul    it's a proxy cache ?
22:19 sofar      just git pull and update mods on request
22:19 nerzhul    okay
22:20 sofar      so I can selectively deny things I don't want to host
22:20 sofar      see: media.foo-projects.org
22:20 sofar      erg, hmm
22:20 sofar      https://forum.minetest.net/viewtopic.php?f=14&t=18951&start=25
22:20 sofar      there
22:21 sofar      erm click on page 1, sorry for bad link
22:22 sofar      nerzhul: I get about 30gb of transfer per month atm, and about 100 player requests a day (3000 unique a month)
22:22 nerzhul    interesting
22:22 nerzhul    maintenance can be huge if minetest population increase
22:23 sofar      the golang implementation I made is doing great and nginx does the heavy transfer work
22:23 sofar      we can always CDN the content, too
22:30 sofar      hmm, maybe we can have curl send referrer headers to remote media server reqests somehow
22:35 sofar      maybe in clientmedia.cpp around line 258?
22:37 oiaohm     sofar: I am reading about Remote Media in minetest what is crossing my mind is git fetch  and git checkout FETCH_HEAD - - <file>    I know it might be horrible idea but the ablity to use github it self as the media server instead of running own.
22:39 sfan5      if you consider that most users are on mobile operating systems, you'll find that none of them have (or even want) git installed
22:39 sofar      git isn't very efficient to get content to stuff like a browser or a game
22:40 sofar      it's efficient for developer workflows
22:40 sfan5      also that, git isn't made for storage of binary files
22:42 sofar      the way it works is really quite nice, the client gets exactly what they need and they can verify it too
22:42 sofar      the server can have 10 different versions of default_grass.png without issues
22:45 oiaohm     sfan5: you are right and wrong on git is not design for storage binaries.    https://git-lfs.github.com/   Lot of git hosting has git-lfs enabled like github and gitlab and that is design to host binaries.
22:45 sfan5      git-lfs is not part of the git "project" as far as I'm aware
22:53 sofar      it's an interesting thought but the need for something like this is nonexistant
22:53 sofar      the remote media protocol is well written and efficient, and can leverage things like pipelining and ssl/tls if wanted even
22:53 sofar      throwing it all out is just a waste of time
22:58 oiaohm     sfan5: correct is not part of the git project but its insanely common extention on most git hosting.
23:00 oiaohm     sofar: as I said what I suggest might be a horrible idea.  git-lfs to function requires https.   If the remote media server could https provide the server it could be used for compacting git clones.  I am not saying through out own remote media server option.   Due mods and so on being common developed in git looking at intergration could be good.
23:01 oiaohm     through/throw
23:05 Calinou    fetching custom game data via HTTP/HTTPS is tried and proven :)
23:05 sofar      I think he's thinking that servers could rely on git to get game data
23:06 sofar      anyway I fail to see any benefit, shrug
23:06 nerzhul    sofar yeah i like your golang code
23:17 sofar      makes deployment easy, but it's also nice and simple
23:17 sofar      found a few little bugs over time, nothing terrible
23:26 nerzhul    merging #8044
23:26 ShadowBot  https://github.com/minetest/minetest/issues/8044 -- Remove remote media compatibility mode. by sofar
23:28 rubenwardy nerzhul: the player backend does, however
23:28 rubenwardy it causes seconds of lag
23:28 rubenwardy every 18 seconds it freezes